Aiming at the problem of sample missing for magnetic flux leakage (MFL), a data reconstruction method based on conditional autoencoder (CVAE) and generative adversarial networks (GAN) is proposed.

This paper proposes a novel collaborative dysarthric speech recognition system designed to convert dysarthric speech into non-dysarthric speech to enhance the robustness of automatic speech ...